Sync API (Single Image)

The Image Moderation API detects risks such as political sensitivity, pornography, advertising, and terrorism in images, and can additionally identify faces, logos, animals, plants, and other content based on the business scenarios enabled on your account.

API Description

Synchronous single image moderation API. Submits one image in a single request and returns a moderation result for each image directly. Detects risks such as political sensitivity, pornography, advertising, and terrorism, and can additionally identify faces, logos, animals, plants, and other content based on the business scenarios enabled on your account.

Requirements

Item	Specification
Protocol	HTTP or HTTPS
Method	POST
Encoding	UTF-8
Format	All request and response parameters use JSON

Image Requirements

Item	Specification
Image types	URL, BASE64
Supported formats	jpg, jpeg, png, webp, gif, tiff, tif, heif, heic, avif, apng, bmp, svg (static only)
Resolution	20×20 px to 6000×6000 px. Recommended minimum: 256×256 px
Size limit	Synchronous: ≤ 10 MB. Asynchronous: ≤ 30 MB

Frame extraction: Tall, scrolling images (long-form screenshots, product detail pages, comic-style posts) are submitted as a single frame by default. Contact DeepCleer to enable automatic splitting into multiple frames — note that this increases the number of billable moderation units per request. Animated images such as GIFs are always split into frames, according to the parameters you pass in the request.

ℹ️
Host images on a highly available CDN before submitting their URLs. DeepCleer fetches each image directly from the origin server you provide, so any outage or single point of failure on your side will cause fetch failures — and an image that can't be fetched can't be moderated.

Timeout Suggestion

Synchronous single image: recommended timeout of 10 seconds
Asynchronous single image: recommended timeout of 5 seconds

ℹ️
End-to-end response time is dominated by how long DeepCleer takes to fetch your image — keep your hosting fast and reliable. Fetch timeouts are 2 seconds to connect and 3 seconds to read, with one automatic retry on failure. Once the image is in hand, DeepCleer's own processing averages about 500 ms, with the exact figure varying by detection type and image size.

Request

Request URL

Cluster	Request URL
Silicon Valley	`http://api-img-gg.fengkongcloud.com/image/v4`
Singapore	`http://api-img-xjp.fengkongcloud.com/image/v4`

Request Parameters

Parameter	Type	Required	Max Length	Description
`accessKey`	string	Yes	20	API authentication key. The default `accessKey` is sent in your onboarding email.
`appId`	string	Yes	64	Application identifier, such as `web` for your web application or `app` for your mobile app. The default `appId` is sent in your onboarding email. Contact DeepCleer if you need a new `appId`.Event identifier used to distinguish moderation scenarios in your application, such as `prompt` for user prompts or `modelOutput` for LLM output. The default `eventId` is sent in your onboarding email. Contact DeepCleer if you need a new `eventId`.
`eventId`	string	Yes	64	Event identifier used to distinguish moderation scenarios in your application, such as uploaded `attachments` of user prompts or `modelOutput` for LLM output. The default eventId is sent in your onboarding email. Contact DeepCleer if you need a new eventId.
`type`	string	No	64	Risk detection types to run. Either `type` or `businessType` must be provided; you can also provide both. Multiple values can be combined with underscores (for example, `BAN_EROTIC`). Detailed risk detection types can refer to Detection Types.
`businessType`	string	No	128	Business detection types to run — your organization's custom moderation categories, configured with DeepCleer separately from the built-in `type` catalog. Either `type` or `businessType` must be provided; you can also provide both. Multiple values can be combined with underscores. See the Business Label Reference for available values.
`data`	object	Yes	—	Request data content. See data Object.
`acceptLang`	string	Yes	—	Language for returned labels. Set `en` by default.

Detection Types

Value	Description
`POLITY`	Political content
`EROTIC`	Pornographic and sexually suggestive content
`VIOLENT`	Violence, terrorism, and other prohibited content
`QRCODE`	QR codes embedded in the image
`ADVERT`	Advertising imagery
`IMGTEXTRISK`	Risky text found inside the image itself (OCR-based — required if your images contain overlaid captions, memes, or screenshots of text)
`BOCR`	Multi-language OCR with automatic language detection.

`data` Object

Parameter	Type	Required	Max Length	Description
`img`	string	Yes	1024	The image to moderate. Accepts either a publicly fetchable URL or a base64-encoded image payload.
`backupUrl`	string	No	1024	Fallback URL for the image. If DeepCleer cannot download `img` (timeout or DNS failure), it retries against this URL once.
`tokenId`	string	Yes	64	Stable identifier for the end user, typically your internal user UID (an encrypted UID is fine). Used for behavioral-risk signals such as spam and repeat-offender detection. Alphanumeric with underscores and hyphens, up to 64 characters.
`dataId`	string	No	128	Client-side identifier attached to the moderation call. DeepCleer echoes it back with the result, letting you correlate your source record (a post ID, message ID, review ID, etc.) with the moderation verdict — typically used to look up historical decisions in your own database or in the DeepCleer console.
`deviceId`	string	No	128	Device-fingerprint identifier issued by DeepCleer.
`gender`	string	No	—	User's gender. `0` male, `1` female, `2` unknown.
`imgCompareBase`	string	No	1024	Reference image used as the comparison target for face matching. Required when `businessType` includes `FACECOMPARE`. Accepts a publicly fetchable URL or a base64-encoded image payload. Supported formats: JPG/JPEG, PNG, WEBP, GIF, TIFF/TIF, HEIF. Image must be at least 256×256 px and no larger than 10 MB. Tall scrolling images and animated images are not accepted as the reference — pick a single, conventionally proportioned still frame showing the target face clearly.
`interval`	int32	No	—	Frame-sampling stride for animated images. Defaults to `1`, which extracts every frame. With a value of N, DeepCleer keeps one frame out of every N. If the resulting frame count would exceed `maxFrame`, DeepCleer widens the stride automatically to `totalFrames` / `maxFrame` so the cap is always respected — meaning `maxFrame` acts as a hard ceiling and `interval` only controls sampling density when there's room under the cap.
`maxFrame`	int32	No	—	Upper bound on how many frames DeepCleer extracts from an animated image. Defaults to `3`; the maximum accepted value is `20`. Works together with `interval` — whichever setting produces fewer frames wins, so `maxFrame` acts as a hard ceiling. Billing is based on the number of frames actually extracted and moderated, not on `maxFrame` itself — a 3-frame GIF billed with `maxFrame`=20 still costs three moderation units.
`ip`	string	No	64	Public IP address of the user who submitted the text. Accepts IPv4 or IPv6.
`lang`	string	Yes	—	Language of the text embedded inside the image, used by the OCR-based text detectors. Set to `en` by default. Supported values: `en` (English), `ar` (Arabic), `zh` (Chinese) and `auto` (automatic language detection — requires `BOCR` in the `type` field.
`level`	int32	No	—	User level. See User Levels.
`nickname`	string	No	150	The end user's display name, moderated alongside text. Up to 150 characters; longer values are truncated
`receiveTokenId`	string	No	64	`tokenId` of the message recipient in a one-to-one chat. Alphanumeric with underscores and hyphens, up to 64 characters. Required when `eventId` is `message`.
`extra`	object	No	—	Auxiliary parameters. See extra Object.
`streamInfo`	object	No	—	Deduplication settings for video-stream moderation results. When a video stream shows the same risk across many consecutive frames (for example, a political banner held on screen for ten seconds), this object controls how those redundant detections are collapsed before they're returned to you — trading notification volume for granularity. See the streamInfo Object for the individual fields.

User Levels

Value	Description
`0`	Lowest-level user (e.g., newly registered, completely inactive, or level-0 users)
`1`	Lower-level user (e.g., low activity or low-level users)
`2`	Mid-level user (e.g., moderately active or mid-level users)
`3`	Higher-level user (e.g., highly active or high-level users)
`4`	Highest-level user (e.g., paying users, VIP users)

`extra` Object

Parameter	Type	Required	Max Length	Description
`isIgnoreTls`	boolean	No	—	Whether to skip CA-certificate verification when DeepCleer fetches `img` or `backupUrl` over HTTPS. Defaults to `false` (verify the certificate). Set to `true` to skip verification — useful for self-signed dev/staging hosts, but never recommended for production traffic.
`isTokenSeparate`	int32	No	—	Whether per-account behavioral signals are scoped to each `appId` separately or pooled across all of your applications. `0` (default) pools signals — a `tokenId`'s history follows the user across every `appId` on your account. `1` isolates them — each `appId` maintains its own independent behavioral history for the same `tokenId`.
`room`	string	No	64	Live-room or game-room identifier. Lets DeepCleer apply room-specific moderation rules and accumulate per-room context for behavioral detectors.
`passThrough`	object	No	1024	Arbitrary pass-through data. The value is echoed back verbatim in the response, unchanged.

`streamInfo` Object

Controls risk return frequency for frames from the same video stream. Useful for reducing manual review costs.

Parameter	Type	Required	Max Length	Description
`streamId`	string	No	64	Stable identifier for the video stream this frame belongs to. DeepCleer uses it to group frames from the same stream so deduplication can apply across them. Required when `similarDedup` is `true`.
`similarDedup`	boolean	No	—	Whether to deduplicate near-identical risk results within the same stream. When enabled, DeepCleer compares each new detection against recent ones from the same `streamId` and suppresses repeats — preventing a single sustained risk (a banner held on screen, a song looping in the background) from generating one callback per frame. Defaults to `false`.
`frameTime`	int32	No	13	Capture time of this frame, as a 13-digit Unix timestamp in milliseconds (UTC). DeepCleer uses it to decide whether two detections fall inside the same `timeWindow`. Required when `similarDedup` is `true`.
`riskNum`	int32	No	—	How many detections of the same risk category to surface per `timeWindow`, per `streamId`. Range `1`–`5`; defaults to `1`. Lower values = quieter callbacks, less timeline detail; higher values = more callbacks, finer-grained timing.
`timeWindow`	int32	No	64	Length of the deduplication window, in seconds. Detections of the same risk category arriving inside this window for the same `streamId` are subject to the `riskNum` cap. Required when `similarDedup` is `true`. Defaults to `10`.

Response

Response Parameters

ℹ️
Parameters other than code, message, and requestId are only guaranteed to be returned when code is 1100.

Parameter	Type	Required	Description
`requestId`	string	Yes	Unique identifier for this request. We strongly recommend logging it on your side — it's the key piece of information DeepCleer needs to help you investigate a specific moderation decision or tune your integration.
`code`	int32	Yes	Response code indicating call status. See Response Codes.
`message`	string	Yes	Human-readable message corresponding to `code`.
`riskLevel`	string	Yes	Disposition recommendation. `PASS` — content is clean; allow it. `REVIEW` — content is suspicious; route to human review. `REJECT` — content violates policy; block it.
`riskLabel1`	string	Yes	Level-1 risk label. Returns `normal` when `riskLevel` is `PASS`; otherwise returns a violation tag such as `politics`, `porn`, or `ad`. See the Risk Label Reference for the full list.
`riskLabel2`	string	Yes	Level-2 risk label, narrowing the category named in `riskLabel1`. Empty when `riskLevel` is `PASS`.
`riskLabel3`	string	Yes	Level-3 risk label, narrowing the category named in `riskLabel2`. Empty when `riskLevel` is `PASS`.
`riskDescription`	string	Yes	Human-readable summary of the moderation verdict, formed by joining the three `riskLabel` levels with colons (for example, `Advertising: Contact information: Contact information`). Returns `Normal` when `riskLevel` is `PASS`. *For display purposes only — do not parse or branch on this string; use the `riskLabel` fields for programmatic logic.**
`resultType`	int32	Yes	Stage of the current result. `0` machine moderation; `1` human moderation.
`finalResult`	int32	Yes	Whether this is the final result. `0` intermediate (machine result; human review will follow); `1` final (safe to apply downstream immediately). Always `1` for accounts configured with machine moderation only.
`riskDetail`	object	Yes	Structured evidence backing the verdict — matched keyword-list entries and high-risk text spans, with positions. See the riskDetail object. Empty when the verdict is purely model-driven.
`auxInfo`	object	Yes	Auxiliary output such as masked text and detected contact information. See the auxInfo object.
`allLabels`	array	Yes	Every risk label that matched, with confidence scores and per-label details — useful when you need finer granularity than the top-level `riskLabel*` fields. See the allLabels array.
`businessLabels`	array	No	All matched business-label results (the fine-grained attribute detectors enabled via `businessType`). See the businessLabels array.
`disposal`	object	No	Result re-mapped onto your in-house label taxonomy. Returned only when a custom label system has been configured with DeepCleer. See the disposal object.
`tokenLabels`	object	Yes*	Account-level risk profile derived from the user's history. Returned only when `tokenId` is provided in the request and the account-labeling service is enabled. See the tokenLabels object.
`tokenProfileLabels`	array	No	Account attribute labels (demographics and stable behavioral traits). Returned only when `tokenId` is provided and the account-labeling service is enabled. See Token Labels.
`tokenRiskLabels`	array	No	Account risk labels (recent suspicious activity, prior violations). Returned only when `tokenId` is provided and the account-labeling service is enabled. See Token Labels.

Response Codes

Code	Meaning
`1100`	Success
`1901`	QPS limit exceeded
`1902`	Invalid request parameters
`1903`	Service-side error
`1911`	Could not download the image (timeout, HTTP error, or DNS failure on `img` and `backupUrl` if provided)
`9101`	Unauthorized — check your `accessKey` and that your account has the requested `type` enabled

`riskDetail` Object

Parameter	Type	Required	Description
`riskSource`	int32	Yes	Where the verdict came from. `1000` no risk; `1001` risk found in text (OCR or accompanying caption); `1002` risk found in the visual content of the image itself.
`face_num`	int32	No	Number of faces detected in the image.
`person_num`	int32	No	Number of people (full bodies) detected in the image.
`faces`	array	No	Identified faces flagged as politically sensitive figures, with names and bounding boxes. Capped at 10 entries — when more than 10 match, only the highest-confidence ones are returned. See the Face object.
`objects`	array	No	Detected objects and brand logos, with names and bounding boxes. See Object info.
`ocrText`	object	No	Text extracted from the image by OCR. Returned when `type` includes `IMGTEXTRISK` or `ADVERT`. Contains a single field `text` (`string`) holding the recognized text.
`matchedLists`	array	No	Custom keyword lists triggered by the OCR-extracted text. See Matched lists.
`riskSegments`	array	No	High-risk text spans extracted from the image. See Risk segments.
`persons`	array	No	Detected people with positions and confidence scores. Capped at 10 entries. See the Person object.

Face object

Parameter	Type	Required	Description
`id`	string	No	Stable identifier for this face occurrence. The same person at the same position keeps the same `id` across all labels in the response; if the same person appears N times in the image (different positions), they receive N separate `id`s.
`name`	string	No	Recognized name of the person.
`face_ratio`	float	No	Fraction of the image area occupied by this face (`0`–`1`).
`probability`	float	No	Confidence score (`0`–`1`). Higher is more confident.
`location`	array	No	Bounding box as `[x1, y1, x2, y2]` — top-left and bottom-right pixel coordinates.

Object info

Parameter	Type	Required	Description
`id`	string	No	Stable identifier for this object/logo occurrence.
`name`	string	No	Recognized name of the object or logo.
`probability`	float	No	Confidence score (`0`–`1`).
`qrContent`	string	No	URL or text decoded from a QR code, when the detected object is a QR code.
`location`	array	No	Bounding box as `[x1, y1, x2, y2]`.

Matched lists

Parameter	Type	Required	Description
`name`	string	No	Name of the matched list, as configured in the DeepCleer console.
`words`	array	No	Individual terms from the list that matched the OCR-extracted text.
`words[].word`	string	No	The matched term, as it appeared on the list.
`words[].position`	array	No	Character offsets `[start, end]` of the match within the OCR text. Zero-indexed; `end` is exclusive.

Risk segments

Parameter	Type	Required	Description
`segment`	string	No	The high-risk text span, extracted verbatim from the OCR output.
`position`	array	No	Character offsets `[start, end]` of the span within the OCR text. Zero-indexed; `end` is exclusive.

Person object

Parameter	Type	Required	Description
`id`	string	No	Stable identifier for this person occurrence — kept consistent across all labels in the response.
`person_ratio`	float	No	Fraction of the image area occupied by this person (`0`–`1`).
`probability`	float	No	Confidence score (`0`–`1`).
`location`	array	No	Bounding box as `[x1, y1, x2, y2]`.

`auxInfo` Object

Parameter	Type	Required	Description
`segments`	int32	Yes	Number of frames actually moderated. For static images this is always `1`; for animated images it reflects the count after `interval` and `maxFrame` have been applied.
`downloadTime`	integer	No	Time DeepCleer spent fetching the image, in milliseconds. Useful for diagnosing slow responses caused by upstream hosting.
`totalProcessTime`	integer	No	Total time DeepCleer spent on this request — fetch plus inference — in milliseconds.
`qrContent`	string	No	URL or text decoded from a QR code found in the image. Mirrors the `qrContent` returned at the object level when QR detection is the dominant signal.
`frameTime`	int32	No	Capture timestamp of the suppressed frame. Returned when `streamInfo` is supplied and similarity deduplication is triggered.
`streamId`	string	No	Stream identifier of the suppressed frame. Returned when `streamInfo` is supplied and similarity deduplication is triggered.
`passThrough`	object	No	Pass-through data echoed back verbatim from `data.extra.passThrough` in the request.

`allLabels` Array

Each element has the following fields:

Parameter	Type	Required	Description
`riskLevel`	string	No	Disposition for this individual label: `PASS`, `REVIEW`, or `REJECT`.
`riskLabel1`	string	No	Level-1 risk label.
`riskLabel2`	string	No	Level-2 risk label, narrowing `riskLabel1`.
`riskLabel3`	string	No	Level-3 risk label, narrowing `riskLabel2`.
`riskDescription`	string	No	Human-readable summary for display. *Do not parse or branch on this string; use the `riskLabel` fields for programmatic logic.**
`probability`	float	No	Confidence score (`0`–`1`). Higher is more confident.
`riskDetail`	object	No	Per-label evidence — same shape as the top-level riskDetail object.

`businessLabels` Array

Each element has the following fields:

Parameter	Type	Required	Description
`businessLabel1`	string	Yes	Level-1 business label.
`businessLabel2`	string	Yes	Level-2 business label, narrowing `businessLabel1`.
`businessLabel3`	string	Yes	Level-3 business label, narrowing `businessLabel2`.
`businessDescription`	string	Yes	Human-readable summary, formed by joining the three `businessLabel` levels with colons (for example, `Object: Vehicle: Sedan`). For display only — use the `businessLabel*` fields for programmatic logic.
`probability`	float	Yes	Raw confidence score (`0`–`1`).
`confidenceLevel`	int32	No	Bucketed confidence (`0`–`2`). Higher values indicate greater confidence — useful when you want a tier ("low/medium/high") instead of a continuous score.
`businessDetail`	object	No	Detector-specific evidence. May contain any of `face_num`, `person_num`, `face_ratio`, `person_ratio`, `face_compare_num`, `name`, `probability`, `faces`, `objects`, `persons`, and `location` — same shapes as the corresponding fields in the riskDetail object. Which fields appear depends on which `businessType` detector produced the label.

`disposal` Object

Returned only when a custom label system has been configured with DeepCleer. DeepCleer can re-map its internal labels onto identifiers in your in-house taxonomy and return results using your naming. Contact your account manager to enable.

Parameter	Type	Required	Description
`riskLevel`	string	Yes	Mapped disposition recommendation. If no rule matches, the default disposition is returned.
`riskLabel1`	string	Yes	Mapped level-1 label. Returns `normal` when there is no mapping and `riskLevel` is `PASS`.
`riskLabel2`	string	Yes	Mapped level-2 label. Empty when there is no mapping and `riskLevel` is `PASS`.
`riskLabel3`	string	Yes	Mapped level-3 label. Empty when there is no mapping and `riskLevel` is `PASS`.
`riskDescription`	string	Yes	Mapped human-readable summary. Returns `Normal` when `riskLevel` is `PASS`.
`riskDetail`	object	Yes	Mapped evidence — same shape as the top-level riskDetail object.

`tokenLabels` Object

Account-level risk profile derived from the user's history across your applications. Returned only when tokenId is provided in the request and the account-labeling service is enabled on your plan.

Parameter	Type	Required	Description
`UGC_account_risk`	object	No	Risks inferred from the user's user-generated-content history.
`UGC_account_risk.b_advertise_risk_tokenid`	int32	No	Advertising-spam risk. `0` no risk found; `1` risk present.
`UGC_account_risk.b_advertise_risk_tokenid_last_ts`	int32	No	Last time the advertising-spam risk was observed. 13-digit Unix timestamp in milliseconds (UTC).
`UGC_account_risk.b_politics_risk_tokenid`	int32	No	Political-content risk. `0` no risk found; `1` risk present.
`UGC_account_risk.b_politics_risk_tokenid_last_ts`	int32	No	Last time the political-content risk was observed. 13-digit Unix timestamp in milliseconds (UTC).
`UGC_account_risk.b_sexy_risk_tokenid`	int32	No	Sexual/pornographic-content risk. `0` no risk found; `1` risk present.
`UGC_account_risk.b_sexy_risk_tokenid_last_ts`	int32	No	Last time the sexual-content risk was observed. 13-digit Unix timestamp in milliseconds (UTC).
`machine_account_risk`	object	No	Risks indicating the account is automated rather than human-operated.
`machine_account_risk.b_machine_control_tokenid`	int32	No	Bot/automation risk. `0` appears human-operated; `1` appears machine-controlled.
`machine_account_risk.b_machine_control_tokenid_last_ts`	int32	No	Last time the bot risk was observed. 13-digit Unix timestamp in milliseconds (UTC).
`machine_account_risk.b_offer_wall_tokenid`	int32	No	Offer-wall (incentivized signup) risk — accounts created or driven by reward-for-action campaigns. `0` not an offer-wall account; `1` is one.
`machine_account_risk.b_offer_wall_tokenid_last_ts`	int32	No	Last time the offer-wall risk was observed. 13-digit Unix timestamp in milliseconds (UTC).
`scene_account_risk`	object	No	Industry-specific account risks (for example, airline ticketing).
`scene_account_risk.i_tout_risk_tokenid`	int32	No	Airline seat-blocking risk — accounts that hold seat reservations to resell. `0` not a seat-blocker; `1` is one.
`scene_account_risk.i_tout_risk_tokenid_last_ts`	int32	No	Last time the seat-blocking risk was observed. 13-digit Unix timestamp in milliseconds (UTC).

Token Labels

tokenProfileLabels and tokenRiskLabels share the same element shape:

Parameter	Type	Required	Description
`label1`	string	No	Level-1 label.
`label2`	string	No	Level-2 label, narrowing `label1`.
`label3`	string	No	Level-3 label, narrowing `label2`.
`description`	string	No	Human-readable description of the label, for display only.
`timestamp`	int32	No	When the label was assigned. 13-digit Unix timestamp in milliseconds (UTC).

Examples

Request Example

{
  "accessKey": "YOUR_ACCESS_KEY",
  "appId": "default",
  "eventId": "default",
  "type": "POLITY_EROTIC_VIOLENT_QRCODE_ADVERT",
  "data": {
    "img": "http://example.com/image.jpg",
    "tokenId": "username123"
  }
}

Response Example

{
  "allLabels": [
    {
      "probability": 0.922851562500734,
      "riskDescription": "Violence: Terrorism scene: Bloody",
      "riskDetail": {
        "ocrText": {
          "text": "Sample OCR text content"
        },
        "riskSource": 1002
      },
      "riskLabel1": "violence",
      "riskLabel2": "baokongchangjing",
      "riskLabel3": "xuexing",
      "riskLevel": "REJECT"
    }
  ],
  "auxInfo": {
    "segments": 1
  },
  "businessLabels": [
    {
      "businessDescription": "Face: Gender: Male",
      "businessDetail": {
        "face_ratio": 0.00106996833346784,
        "faces": [
          {
            "face_ratio": 0.00106996833346784,
            "id": "35da33429897372017bb7ad7c7302693",
            "location": [228, 588, 256, 622],
            "name": "",
            "probability": 0.967536687850952
          }
        ],
        "location": [228, 588, 256, 622],
        "name": "",
        "probability": 0.967536687850952
      },
      "businessLabel1": "face",
      "businessLabel2": "gender",
      "businessLabel3": "male",
      "confidenceLevel": 2,
      "probability": 0.967536687850952
    },
    {
      "businessDescription": "Face: Face type: Real person",
      "businessDetail": {
        "face_num": 1,
        "face_ratio": 0.00106996833346784,
        "faces": [
          {
            "face_ratio": 0.00106996833346784,
            "id": "35da33429897372017bb7ad7c7302693",
            "location": [228, 588, 256, 622],
            "name": "Example Person",
            "probability": 0.541935835649002
          }
        ],
        "location": [228, 588, 256, 622],
        "name": "Example Person",
        "probability": 0.541935835649002
      },
      "businessLabel1": "face",
      "businessLabel2": "renlianleixing",
      "businessLabel3": "zhenren",
      "confidenceLevel": 2,
      "probability": 0.996093758527634
    },
    {
      "businessDescription": "Face: Face pose: Frontal face",
      "businessDetail": {},
      "businessLabel1": "face",
      "businessLabel2": "renlianzitai",
      "businessLabel3": "zhenglian",
      "confidenceLevel": 1,
      "probability": 0.450656906102068
    }
  ],
  "code": 1100,
  "finalResult": 1,
  "message": "Success",
  "requestId": "7cf340c0910b768e6904b342b908a4f0",
  "resultType": 0,
  "riskDescription": "Violence: Terrorism scene: Bloody",
  "riskDetail": {
    "ocrText": {
      "text": "Sample OCR text content"
    },
    "riskSource": 1002
  },
  "riskLabel1": "violence",
  "riskLabel2": "baokongchangjing",
  "riskLabel3": "xuexing",
  "riskLevel": "REJECT",
  "tokenLabels": {
    "UGC_account_risk": {}
  }
}

API Description

Requirements

Image Requirements

Timeout Suggestion

Request

Request URL

Request Parameters

Detection Types

data Object

User Levels

extra Object

streamInfo Object

Response

Response Parameters

Response Codes

riskDetail Object

Face object

Object info

Matched lists

Risk segments

Person object

auxInfo Object

allLabels Array

businessLabels Array

disposal Object

tokenLabels Object

Token Labels

Examples

Request Example

Response Example

`data` Object

`extra` Object

`streamInfo` Object

`riskDetail` Object

`auxInfo` Object

`allLabels` Array

`businessLabels` Array

`disposal` Object

`tokenLabels` Object